Large-margin structured prediction via linear programming

Machine Translation, 2010

We propose a structured learning approach, max-margin structure (MMS), which is targeted at natural language processing (NLP) tasks. The architecture of our approach is shown to capture structural aspects of the problem domains, leading to demonstrable performance improvements on two NLP tasks: part-of-speech tagging and statistical machine translation (SMT). We present a perceptron-based online learning algorithm to train the model and demonstrate desirable computational scaling behavior over traditional optimisation methods.

ABSTRACT We propose an ecient discriminative train- ing method for generative models under su- pervised learning. In our setting, fully ob- served instances are given as training exam- ples, together with a specification of vari- ables of interest for prediction. We formu- late the training as a convex programming problem, incorporating the SVM-type large margin constraints to favor parameters under which the maximum a posteriori (MAP) esti- mates of the prediction variables, conditioned on the rest, are close to their true values given in the training instances. The resulting opti- mization problem is, however, more complex than its quadratic programming (QP) coun- terpart resulting from the SVM-type training of conditional models, because of the pres- ence of non-linear constraints on the param- eters. We present an ecient optimization method, which combines several techniques, namely, a data-dependent reparametrization of dual variables, restricted simplicial de- composition, and the proximal point algo- rithm. Our method extends the one for solv- ing the aforementioned QP counterpart, pro- posed earlier by some of the authors.

2012

Abstract: Structured prediction is the cornerstone of several machine learning applications. Unfortunately, in structured prediction settings with expressive inter-variable interactions, exact inference-based learning algorithms, eg Structural SVM, are often intractable. We present a new way, Decomposed Learning (DecL), which performs efficient learning by restricting the inference step to a limited part of the structured spaces.

Journal of Machine Learning Research, 2014

We consider a framework for structured prediction based on search in the space of complete structured outputs. Given a structured input, an output is produced by running a timebounded search procedure guided by a learned cost function, and then returning the least cost output uncovered during the search. This framework can be instantiated for a wide range of search spaces and search procedures, and easily incorporates arbitrary structuredprediction loss functions. In this paper, we make two main technical contributions. First, we describe a novel approach to automatically defining an effective search space over structured outputs, which is able to leverage the availability of powerful classification learning algorithms. In particular, we define the limited-discrepancy search space and relate the quality of that space to the quality of learned classifiers. We also define a sparse version of the search space to improve the efficiency of our overall approach. Second, we give a generic cost function learning approach that is applicable to a wide range of search procedures. The key idea is to learn a cost function that attempts to mimic the behavior of conducting searches guided by the true loss function. Our experiments on six benchmark domains show that a small amount of search in limited discrepancy search space is often sufficient for significantly improving on state-of-the-art structured-prediction performance. We also demonstrate significant speed improvements for our approach using sparse search spaces with little or no loss in accuracy.

2012

We consider a framework for structured prediction based on search in the space of complete structured outputs. Given a structured input, an output is produced by running a time-bounded search procedure guided by a learned cost function, and then returning the least cost output uncovered during the search. This framework can be instantiated for a wide range of search spaces and search procedures, and easily incorporates arbitrary structured-prediction loss functions. In this paper, we make two main technical contributions. First, we define the limited-discrepancy search space over structured outputs, which is able to leverage powerful classification learning algorithms to improve the search space quality. Second, we give a generic cost function learning approach, where the key idea is to learn a cost function that attempts to mimic the behavior of conducting searches guided by the true loss function. Our experiments on six benchmark domains demonstrate that using our framework with only a small amount of search is sufficient for significantly improving on state-of-the-art structuredprediction performance.

A common obstacle preventing the rapid deployment of supervised machine learning algorithms is the lack of labeled training data. This is particularly expensive to obtain for structured prediction tasks, where each training instance may have multiple, interacting labels, all of which must be correctly annotated for the instance to be of use to the learner. Traditional active learning addresses this problem by optimizing the order in which the examples are labeled to increase learning efficiency. However, this approach does not consider the difficulty of labeling each example, which can vary widely in structured prediction tasks. For example, the labeling predicted by a partially trained system may be easier to correct for some instances than for others. We propose a new active learning paradigm which reduces not only how many instances the annotator must label, but also how difficult each instance is to annotate. The system leverages information from partially correct predictions to efficiently solicit annotations from the user. We validate this active learning framework in an interactive information extraction system, reducing the total number of annotation actions by 22%.

2009 IEEE International Workshop on Machine Learning for Signal Processing, 2009

We applied a structure learning model, Max-Margin Structure (MMS), to natural language processing (NLP) tasks, where the aim is to capture the latent relationships within the output language domain. We formulate this model as an extension of multi-class Support Vector Machine (SVM) and present a perceptron-based learning approach to solve the problem. Experiments are carried out on two related NLP tasks: part-of-speech (POS) tagging and machine translation (MT), illustrating the effectiveness of the model.

2007

We propose in this chapter a modification of CRF-like algorithms that allows for solving large-scale structured classification problems. Our approach consists in upper bounding the CRF functional in order to decompose its training into independent optimisation problems per clique. Furthermore we show that each sub-problem corresponds to solving a multiclass learning task in each clique, which enlarges the applicability of these tools for large-scale structural learning problems. Before presenting the Conditional Graphical Model (CGM), as we refer to this procedure, we review the family of CRF algorithms. We concentrate on the best known procedures and standard generalisations of CRFs. The objective of this section is analysing from the same viewpoint the proposed solutions in the literature to tackle this problem, which allows comparing their different features. We complete the chapter with a case study, in which we show the possibility to work with largescale problems using CGM and that the obtained performance is comparable to the result with CRF-like algorithms.

2010

Abstract: We present a unified framework for online learning of structured classifiers that handles a wide family of convex loss functions, properly including CRFs, structured SVMs, and the structured perceptron. We introduce a new aggressive online algorithm that optimizes any loss in this family. For the structured hinge loss, this algorithm reduces to 1-best MIRA; in general, it can be regarded as a dual coordinate ascent algorithm. The approximate inference scenario is also addressed.

2010

In discriminative machine learning one is interested in training a system to optimize a certain desired measure of performance, or loss. In binary classification one typically tries to minimizes the error rate. But in structured prediction each task often has its own measure of performance such as the BLEU score in machine translation or the intersection-over-union score in PASCAL segmentation. The most common approaches to structured prediction, structural SVMs and CRFs, do not minimize the task loss: the former minimizes a surrogate loss with no guarantees for task loss and the latter minimizes log loss independent of task loss. The main contribution of this paper is a theorem stating that a certain perceptronlike learning rule, involving features vectors derived from loss-adjusted inference, directly corresponds to the gradient of task loss. We give empirical results on phonetic alignment of a standard test set from the TIMIT corpus, which surpasses all previously reported results on this problem.